The main sociological, philosophical and historical approaches only as
cribe a relative importance to the role of chance, error, or accident
in scientific progress. The literature on this topic tends to be anecd
otal, sometimes hagiographic and rarely systematic. The main goal of t
his paper is to introduce a new approach to the study of serendipity i
n scientific discovery. This new approach is based in the study of hig
hly cited papers obtained from the Citation Classics feature of Curren
t Contents. This paper re-examines 205 Citation Classics commentaries
from the 400 most-cited papers in the recent history of science. Autho
rs of 17 Citation Classics commentaries (8.3%) mention some kind of se
rendipity in performing the research reported in the highly cited pape
r. Commentaries are classified and discussed in detail. In addition, I
have examinated the original papers identified above. In 5 from the o
riginal highly cited papers authors explained or gave enough hints on
the way the serendipitous discovery was done.